1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
19
from bzrlib.lazy_import import lazy_import
20
lazy_import(globals(), """
21
from binascii import hexlify
22
from copy import deepcopy
41
revision as _mod_revision,
50
from bzrlib.osutils import (
55
from bzrlib.revisiontree import RevisionTree
56
from bzrlib.store.versioned import VersionedFileStore
57
from bzrlib.store.text import TextStore
58
from bzrlib.testament import Testament
61
from bzrlib.decorators import needs_read_lock, needs_write_lock
62
from bzrlib.inter import InterObject
63
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
64
from bzrlib.symbol_versioning import (
68
from bzrlib.trace import mutter, note, warning
71
# Old formats display a warning, but only once
72
_deprecation_warning_done = False
75
class Repository(object):
76
"""Repository holding history for one or more branches.
78
The repository holds and retrieves historical information including
79
revisions and file history. It's normally accessed only by the Branch,
80
which views a particular line of development through that history.
82
The Repository builds on top of Stores and a Transport, which respectively
83
describe the disk data format and the way of accessing the (possibly
87
_file_ids_altered_regex = lazy_regex.lazy_compile(
88
r'file_id="(?P<file_id>[^"]+)"'
89
r'.*revision="(?P<revision_id>[^"]+)"'
93
def add_inventory(self, revid, inv, parents):
94
"""Add the inventory inv to the repository as revid.
96
:param parents: The revision ids of the parents that revid
97
is known to have and are in the repository already.
99
returns the sha1 of the serialized inventory.
101
_mod_revision.check_not_reserved_id(revid)
102
assert inv.revision_id is None or inv.revision_id == revid, \
103
"Mismatch between inventory revision" \
104
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
105
assert inv.root is not None
106
inv_text = self.serialise_inventory(inv)
107
inv_sha1 = osutils.sha_string(inv_text)
108
inv_vf = self.control_weaves.get_weave('inventory',
109
self.get_transaction())
110
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
113
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
115
for parent in parents:
117
final_parents.append(parent)
119
inv_vf.add_lines(revid, final_parents, lines)
122
def add_revision(self, rev_id, rev, inv=None, config=None):
123
"""Add rev to the revision store as rev_id.
125
:param rev_id: the revision id to use.
126
:param rev: The revision object.
127
:param inv: The inventory for the revision. if None, it will be looked
128
up in the inventory storer
129
:param config: If None no digital signature will be created.
130
If supplied its signature_needed method will be used
131
to determine if a signature should be made.
133
_mod_revision.check_not_reserved_id(rev_id)
134
if config is not None and config.signature_needed():
136
inv = self.get_inventory(rev_id)
137
plaintext = Testament(rev, inv).as_short_text()
138
self.store_revision_signature(
139
gpg.GPGStrategy(config), plaintext, rev_id)
140
if not rev_id in self.get_inventory_weave():
142
raise errors.WeaveRevisionNotPresent(rev_id,
143
self.get_inventory_weave())
145
# yes, this is not suitable for adding with ghosts.
146
self.add_inventory(rev_id, inv, rev.parent_ids)
147
self._revision_store.add_revision(rev, self.get_transaction())
150
def _all_possible_ids(self):
151
"""Return all the possible revisions that we could find."""
152
return self.get_inventory_weave().versions()
154
def all_revision_ids(self):
155
"""Returns a list of all the revision ids in the repository.
157
This is deprecated because code should generally work on the graph
158
reachable from a particular revision, and ignore any other revisions
159
that might be present. There is no direct replacement method.
161
return self._all_revision_ids()
164
def _all_revision_ids(self):
165
"""Returns a list of all the revision ids in the repository.
167
These are in as much topological order as the underlying store can
168
present: for weaves ghosts may lead to a lack of correctness until
169
the reweave updates the parents list.
171
if self._revision_store.text_store.listable():
172
return self._revision_store.all_revision_ids(self.get_transaction())
173
result = self._all_possible_ids()
174
return self._eliminate_revisions_not_present(result)
176
def break_lock(self):
177
"""Break a lock if one is present from another instance.
179
Uses the ui factory to ask for confirmation if the lock may be from
182
self.control_files.break_lock()
185
def _eliminate_revisions_not_present(self, revision_ids):
186
"""Check every revision id in revision_ids to see if we have it.
188
Returns a set of the present revisions.
191
for id in revision_ids:
192
if self.has_revision(id):
197
def create(a_bzrdir):
198
"""Construct the current default format repository in a_bzrdir."""
199
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
201
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
202
"""instantiate a Repository.
204
:param _format: The format of the repository on disk.
205
:param a_bzrdir: The BzrDir of the repository.
207
In the future we will have a single api for all stores for
208
getting file texts, inventories and revisions, then
209
this construct will accept instances of those things.
211
super(Repository, self).__init__()
212
self._format = _format
213
# the following are part of the public API for Repository:
214
self.bzrdir = a_bzrdir
215
self.control_files = control_files
216
self._revision_store = _revision_store
217
self.text_store = text_store
218
# backwards compatibility
219
self.weave_store = text_store
220
# not right yet - should be more semantically clear ?
222
self.control_store = control_store
223
self.control_weaves = control_store
224
# TODO: make sure to construct the right store classes, etc, depending
225
# on whether escaping is required.
226
self._warn_if_deprecated()
227
self._serializer = xml5.serializer_v5
230
return '%s(%r)' % (self.__class__.__name__,
231
self.bzrdir.transport.base)
234
return self.control_files.is_locked()
236
def lock_write(self):
237
self.control_files.lock_write()
240
self.control_files.lock_read()
242
def get_physical_lock_status(self):
243
return self.control_files.get_physical_lock_status()
246
def missing_revision_ids(self, other, revision_id=None):
247
"""Return the revision ids that other has that this does not.
249
These are returned in topological order.
251
revision_id: only return revision ids included by revision_id.
253
return InterRepository.get(other, self).missing_revision_ids(revision_id)
257
"""Open the repository rooted at base.
259
For instance, if the repository is at URL/.bzr/repository,
260
Repository.open(URL) -> a Repository instance.
262
control = bzrdir.BzrDir.open(base)
263
return control.open_repository()
265
def copy_content_into(self, destination, revision_id=None, basis=None):
266
"""Make a complete copy of the content in self into destination.
268
This is a destructive operation! Do not use it on existing
271
return InterRepository.get(self, destination).copy_content(revision_id, basis)
273
def fetch(self, source, revision_id=None, pb=None):
274
"""Fetch the content required to construct revision_id from source.
276
If revision_id is None all content is copied.
278
return InterRepository.get(source, self).fetch(revision_id=revision_id,
281
def get_commit_builder(self, branch, parents, config, timestamp=None,
282
timezone=None, committer=None, revprops=None,
284
"""Obtain a CommitBuilder for this repository.
286
:param branch: Branch to commit to.
287
:param parents: Revision ids of the parents of the new revision.
288
:param config: Configuration to use.
289
:param timestamp: Optional timestamp recorded for commit.
290
:param timezone: Optional timezone for timestamp.
291
:param committer: Optional committer to set for commit.
292
:param revprops: Optional dictionary of revision properties.
293
:param revision_id: Optional revision id.
295
return _CommitBuilder(self, parents, config, timestamp, timezone,
296
committer, revprops, revision_id)
299
self.control_files.unlock()
302
def clone(self, a_bzrdir, revision_id=None, basis=None):
303
"""Clone this repository into a_bzrdir using the current format.
305
Currently no check is made that the format of this repository and
306
the bzrdir format are compatible. FIXME RBC 20060201.
308
if not isinstance(a_bzrdir._format, self.bzrdir._format.__class__):
309
# use target default format.
310
result = a_bzrdir.create_repository()
311
# FIXME RBC 20060209 split out the repository type to avoid this check ?
312
elif isinstance(a_bzrdir._format,
313
(bzrdir.BzrDirFormat4,
314
bzrdir.BzrDirFormat5,
315
bzrdir.BzrDirFormat6)):
316
result = a_bzrdir.open_repository()
318
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
319
self.copy_content_into(result, revision_id, basis)
323
def has_revision(self, revision_id):
324
"""True if this repository has a copy of the revision."""
325
return self._revision_store.has_revision_id(revision_id,
326
self.get_transaction())
329
def get_revision_reconcile(self, revision_id):
330
"""'reconcile' helper routine that allows access to a revision always.
332
This variant of get_revision does not cross check the weave graph
333
against the revision one as get_revision does: but it should only
334
be used by reconcile, or reconcile-alike commands that are correcting
335
or testing the revision graph.
337
if not revision_id or not isinstance(revision_id, basestring):
338
raise errors.InvalidRevisionId(revision_id=revision_id,
340
return self._revision_store.get_revisions([revision_id],
341
self.get_transaction())[0]
343
def get_revisions(self, revision_ids):
344
return self._revision_store.get_revisions(revision_ids,
345
self.get_transaction())
348
def get_revision_xml(self, revision_id):
349
rev = self.get_revision(revision_id)
351
# the current serializer..
352
self._revision_store._serializer.write_revision(rev, rev_tmp)
354
return rev_tmp.getvalue()
357
def get_revision(self, revision_id):
358
"""Return the Revision object for a named revision"""
359
r = self.get_revision_reconcile(revision_id)
360
# weave corruption can lead to absent revision markers that should be
362
# the following test is reasonably cheap (it needs a single weave read)
363
# and the weave is cached in read transactions. In write transactions
364
# it is not cached but typically we only read a small number of
365
# revisions. For knits when they are introduced we will probably want
366
# to ensure that caching write transactions are in use.
367
inv = self.get_inventory_weave()
368
self._check_revision_parents(r, inv)
372
def get_deltas_for_revisions(self, revisions):
373
"""Produce a generator of revision deltas.
375
Note that the input is a sequence of REVISIONS, not revision_ids.
376
Trees will be held in memory until the generator exits.
377
Each delta is relative to the revision's lefthand predecessor.
379
required_trees = set()
380
for revision in revisions:
381
required_trees.add(revision.revision_id)
382
required_trees.update(revision.parent_ids[:1])
383
trees = dict((t.get_revision_id(), t) for
384
t in self.revision_trees(required_trees))
385
for revision in revisions:
386
if not revision.parent_ids:
387
old_tree = self.revision_tree(None)
389
old_tree = trees[revision.parent_ids[0]]
390
yield trees[revision.revision_id].changes_from(old_tree)
393
def get_revision_delta(self, revision_id):
394
"""Return the delta for one revision.
396
The delta is relative to the left-hand predecessor of the
399
r = self.get_revision(revision_id)
400
return list(self.get_deltas_for_revisions([r]))[0]
402
def _check_revision_parents(self, revision, inventory):
403
"""Private to Repository and Fetch.
405
This checks the parentage of revision in an inventory weave for
406
consistency and is only applicable to inventory-weave-for-ancestry
407
using repository formats & fetchers.
409
weave_parents = inventory.get_parents(revision.revision_id)
410
weave_names = inventory.versions()
411
for parent_id in revision.parent_ids:
412
if parent_id in weave_names:
413
# this parent must not be a ghost.
414
if not parent_id in weave_parents:
416
raise errors.CorruptRepository(self)
419
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
420
signature = gpg_strategy.sign(plaintext)
421
self._revision_store.add_revision_signature_text(revision_id,
423
self.get_transaction())
425
def fileids_altered_by_revision_ids(self, revision_ids):
426
"""Find the file ids and versions affected by revisions.
428
:param revisions: an iterable containing revision ids.
429
:return: a dictionary mapping altered file-ids to an iterable of
430
revision_ids. Each altered file-ids has the exact revision_ids that
431
altered it listed explicitly.
433
assert self._serializer.support_altered_by_hack, \
434
("fileids_altered_by_revision_ids only supported for branches "
435
"which store inventory as unnested xml, not on %r" % self)
436
selected_revision_ids = set(revision_ids)
437
w = self.get_inventory_weave()
440
# this code needs to read every new line in every inventory for the
441
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
442
# not present in one of those inventories is unnecessary but not
443
# harmful because we are filtering by the revision id marker in the
444
# inventory lines : we only select file ids altered in one of those
445
# revisions. We don't need to see all lines in the inventory because
446
# only those added in an inventory in rev X can contain a revision=X
448
unescape_revid_cache = {}
449
unescape_fileid_cache = {}
451
# jam 20061218 In a big fetch, this handles hundreds of thousands
452
# of lines, so it has had a lot of inlining and optimizing done.
453
# Sorry that it is a little bit messy.
454
# Move several functions to be local variables, since this is a long
456
search = self._file_ids_altered_regex.search
457
unescape = _unescape_xml
458
setdefault = result.setdefault
459
pb = ui.ui_factory.nested_progress_bar()
461
for line in w.iter_lines_added_or_present_in_versions(
462
selected_revision_ids, pb=pb):
466
# One call to match.group() returning multiple items is quite a
467
# bit faster than 2 calls to match.group() each returning 1
468
file_id, revision_id = match.group('file_id', 'revision_id')
470
# Inlining the cache lookups helps a lot when you make 170,000
471
# lines and 350k ids, versus 8.4 unique ids.
472
# Using a cache helps in 2 ways:
473
# 1) Avoids unnecessary decoding calls
474
# 2) Re-uses cached strings, which helps in future set and
476
# (2) is enough that removing encoding entirely along with
477
# the cache (so we are using plain strings) results in no
478
# performance improvement.
480
revision_id = unescape_revid_cache[revision_id]
482
unescaped = unescape(revision_id)
483
unescape_revid_cache[revision_id] = unescaped
484
revision_id = unescaped
486
if revision_id in selected_revision_ids:
488
file_id = unescape_fileid_cache[file_id]
490
unescaped = unescape(file_id)
491
unescape_fileid_cache[file_id] = unescaped
493
setdefault(file_id, set()).add(revision_id)
499
def get_inventory_weave(self):
500
return self.control_weaves.get_weave('inventory',
501
self.get_transaction())
504
def get_inventory(self, revision_id):
505
"""Get Inventory object by hash."""
506
return self.deserialise_inventory(
507
revision_id, self.get_inventory_xml(revision_id))
509
def deserialise_inventory(self, revision_id, xml):
510
"""Transform the xml into an inventory object.
512
:param revision_id: The expected revision id of the inventory.
513
:param xml: A serialised inventory.
515
result = self._serializer.read_inventory_from_string(xml)
516
result.root.revision = revision_id
519
def serialise_inventory(self, inv):
520
return self._serializer.write_inventory_to_string(inv)
523
def get_inventory_xml(self, revision_id):
524
"""Get inventory XML as a file object."""
526
assert isinstance(revision_id, basestring), type(revision_id)
527
iw = self.get_inventory_weave()
528
return iw.get_text(revision_id)
530
raise errors.HistoryMissing(self, 'inventory', revision_id)
533
def get_inventory_sha1(self, revision_id):
534
"""Return the sha1 hash of the inventory entry
536
return self.get_revision(revision_id).inventory_sha1
539
def get_revision_graph(self, revision_id=None):
540
"""Return a dictionary containing the revision graph.
542
:param revision_id: The revision_id to get a graph from. If None, then
543
the entire revision graph is returned. This is a deprecated mode of
544
operation and will be removed in the future.
545
:return: a dictionary of revision_id->revision_parents_list.
547
# special case NULL_REVISION
548
if revision_id == _mod_revision.NULL_REVISION:
550
a_weave = self.get_inventory_weave()
551
all_revisions = self._eliminate_revisions_not_present(
553
entire_graph = dict([(node, a_weave.get_parents(node)) for
554
node in all_revisions])
555
if revision_id is None:
557
elif revision_id not in entire_graph:
558
raise errors.NoSuchRevision(self, revision_id)
560
# add what can be reached from revision_id
562
pending = set([revision_id])
563
while len(pending) > 0:
565
result[node] = entire_graph[node]
566
for revision_id in result[node]:
567
if revision_id not in result:
568
pending.add(revision_id)
572
def get_revision_graph_with_ghosts(self, revision_ids=None):
573
"""Return a graph of the revisions with ghosts marked as applicable.
575
:param revision_ids: an iterable of revisions to graph or None for all.
576
:return: a Graph object with the graph reachable from revision_ids.
578
result = graph.Graph()
580
pending = set(self.all_revision_ids())
583
pending = set(revision_ids)
584
# special case NULL_REVISION
585
if _mod_revision.NULL_REVISION in pending:
586
pending.remove(_mod_revision.NULL_REVISION)
587
required = set(pending)
590
revision_id = pending.pop()
592
rev = self.get_revision(revision_id)
593
except errors.NoSuchRevision:
594
if revision_id in required:
597
result.add_ghost(revision_id)
599
for parent_id in rev.parent_ids:
600
# is this queued or done ?
601
if (parent_id not in pending and
602
parent_id not in done):
604
pending.add(parent_id)
605
result.add_node(revision_id, rev.parent_ids)
606
done.add(revision_id)
610
def get_revision_inventory(self, revision_id):
611
"""Return inventory of a past revision."""
612
# TODO: Unify this with get_inventory()
613
# bzr 0.0.6 and later imposes the constraint that the inventory_id
614
# must be the same as its revision, so this is trivial.
615
if revision_id is None:
616
# This does not make sense: if there is no revision,
617
# then it is the current tree inventory surely ?!
618
# and thus get_root_id() is something that looks at the last
619
# commit on the branch, and the get_root_id is an inventory check.
620
raise NotImplementedError
621
# return Inventory(self.get_root_id())
623
return self.get_inventory(revision_id)
627
"""Return True if this repository is flagged as a shared repository."""
628
raise NotImplementedError(self.is_shared)
631
def reconcile(self, other=None, thorough=False):
632
"""Reconcile this repository."""
633
from bzrlib.reconcile import RepoReconciler
634
reconciler = RepoReconciler(self, thorough=thorough)
635
reconciler.reconcile()
639
def revision_tree(self, revision_id):
640
"""Return Tree for a revision on this branch.
642
`revision_id` may be None for the empty tree revision.
644
# TODO: refactor this to use an existing revision object
645
# so we don't need to read it in twice.
646
if revision_id is None or revision_id == _mod_revision.NULL_REVISION:
647
return RevisionTree(self, Inventory(root_id=None),
648
_mod_revision.NULL_REVISION)
650
inv = self.get_revision_inventory(revision_id)
651
return RevisionTree(self, inv, revision_id)
654
def revision_trees(self, revision_ids):
655
"""Return Tree for a revision on this branch.
657
`revision_id` may not be None or 'null:'"""
658
assert None not in revision_ids
659
assert _mod_revision.NULL_REVISION not in revision_ids
660
texts = self.get_inventory_weave().get_texts(revision_ids)
661
for text, revision_id in zip(texts, revision_ids):
662
inv = self.deserialise_inventory(revision_id, text)
663
yield RevisionTree(self, inv, revision_id)
666
def get_ancestry(self, revision_id):
667
"""Return a list of revision-ids integrated by a revision.
669
The first element of the list is always None, indicating the origin
670
revision. This might change when we have history horizons, or
671
perhaps we should have a new API.
673
This is topologically sorted.
675
if revision_id is None:
677
if not self.has_revision(revision_id):
678
raise errors.NoSuchRevision(self, revision_id)
679
w = self.get_inventory_weave()
680
candidates = w.get_ancestry(revision_id)
681
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
684
def print_file(self, file, revision_id):
685
"""Print `file` to stdout.
687
FIXME RBC 20060125 as John Meinel points out this is a bad api
688
- it writes to stdout, it assumes that that is valid etc. Fix
689
by creating a new more flexible convenience function.
691
tree = self.revision_tree(revision_id)
692
# use inventory as it was in that revision
693
file_id = tree.inventory.path2id(file)
695
# TODO: jam 20060427 Write a test for this code path
696
# it had a bug in it, and was raising the wrong
698
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
699
tree.print_file(file_id)
701
def get_transaction(self):
702
return self.control_files.get_transaction()
704
def revision_parents(self, revid):
705
return self.get_inventory_weave().parent_names(revid)
708
def set_make_working_trees(self, new_value):
709
"""Set the policy flag for making working trees when creating branches.
711
This only applies to branches that use this repository.
713
The default is 'True'.
714
:param new_value: True to restore the default, False to disable making
717
raise NotImplementedError(self.set_make_working_trees)
719
def make_working_trees(self):
720
"""Returns the policy for making working trees on new branches."""
721
raise NotImplementedError(self.make_working_trees)
724
def sign_revision(self, revision_id, gpg_strategy):
725
plaintext = Testament.from_revision(self, revision_id).as_short_text()
726
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
729
def has_signature_for_revision_id(self, revision_id):
730
"""Query for a revision signature for revision_id in the repository."""
731
return self._revision_store.has_signature(revision_id,
732
self.get_transaction())
735
def get_signature_text(self, revision_id):
736
"""Return the text for a signature."""
737
return self._revision_store.get_signature_text(revision_id,
738
self.get_transaction())
741
def check(self, revision_ids):
742
"""Check consistency of all history of given revision_ids.
744
Different repository implementations should override _check().
746
:param revision_ids: A non-empty list of revision_ids whose ancestry
747
will be checked. Typically the last revision_id of a branch.
750
raise ValueError("revision_ids must be non-empty in %s.check"
752
return self._check(revision_ids)
754
def _check(self, revision_ids):
755
result = check.Check(self)
759
def _warn_if_deprecated(self):
760
global _deprecation_warning_done
761
if _deprecation_warning_done:
763
_deprecation_warning_done = True
764
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
765
% (self._format, self.bzrdir.transport.base))
767
def supports_rich_root(self):
768
return self._format.rich_root_data
770
def _check_ascii_revisionid(self, revision_id, method):
771
"""Private helper for ascii-only repositories."""
772
# weave repositories refuse to store revisionids that are non-ascii.
773
if revision_id is not None:
774
# weaves require ascii revision ids.
775
if isinstance(revision_id, unicode):
777
revision_id.encode('ascii')
778
except UnicodeEncodeError:
779
raise errors.NonAsciiRevisionId(method, self)
782
class AllInOneRepository(Repository):
783
"""Legacy support - the repository behaviour for all-in-one branches."""
785
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
786
# we reuse one control files instance.
787
dir_mode = a_bzrdir._control_files._dir_mode
788
file_mode = a_bzrdir._control_files._file_mode
790
def get_store(name, compressed=True, prefixed=False):
791
# FIXME: This approach of assuming stores are all entirely compressed
792
# or entirely uncompressed is tidy, but breaks upgrade from
793
# some existing branches where there's a mixture; we probably
794
# still want the option to look for both.
795
relpath = a_bzrdir._control_files._escape(name)
796
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
797
prefixed=prefixed, compressed=compressed,
800
#if self._transport.should_cache():
801
# cache_path = os.path.join(self.cache_root, name)
802
# os.mkdir(cache_path)
803
# store = bzrlib.store.CachedStore(store, cache_path)
806
# not broken out yet because the controlweaves|inventory_store
807
# and text_store | weave_store bits are still different.
808
if isinstance(_format, RepositoryFormat4):
809
# cannot remove these - there is still no consistent api
810
# which allows access to this old info.
811
self.inventory_store = get_store('inventory-store')
812
text_store = get_store('text-store')
813
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
815
def get_commit_builder(self, branch, parents, config, timestamp=None,
816
timezone=None, committer=None, revprops=None,
818
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
819
return Repository.get_commit_builder(self, branch, parents, config,
820
timestamp, timezone, committer, revprops, revision_id)
824
"""AllInOne repositories cannot be shared."""
828
def set_make_working_trees(self, new_value):
829
"""Set the policy flag for making working trees when creating branches.
831
This only applies to branches that use this repository.
833
The default is 'True'.
834
:param new_value: True to restore the default, False to disable making
837
raise NotImplementedError(self.set_make_working_trees)
839
def make_working_trees(self):
840
"""Returns the policy for making working trees on new branches."""
844
def install_revision(repository, rev, revision_tree):
845
"""Install all revision data into a repository."""
848
for p_id in rev.parent_ids:
849
if repository.has_revision(p_id):
850
present_parents.append(p_id)
851
parent_trees[p_id] = repository.revision_tree(p_id)
853
parent_trees[p_id] = repository.revision_tree(None)
855
inv = revision_tree.inventory
856
entries = inv.iter_entries()
857
# backwards compatability hack: skip the root id.
858
if not repository.supports_rich_root():
859
path, root = entries.next()
860
if root.revision != rev.revision_id:
861
raise errors.IncompatibleRevision(repr(repository))
862
# Add the texts that are not already present
863
for path, ie in entries:
864
w = repository.weave_store.get_weave_or_empty(ie.file_id,
865
repository.get_transaction())
866
if ie.revision not in w:
868
# FIXME: TODO: The following loop *may* be overlapping/duplicate
869
# with InventoryEntry.find_previous_heads(). if it is, then there
870
# is a latent bug here where the parents may have ancestors of each
872
for revision, tree in parent_trees.iteritems():
873
if ie.file_id not in tree:
875
parent_id = tree.inventory[ie.file_id].revision
876
if parent_id in text_parents:
878
text_parents.append(parent_id)
880
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
881
repository.get_transaction())
882
lines = revision_tree.get_file(ie.file_id).readlines()
883
vfile.add_lines(rev.revision_id, text_parents, lines)
885
# install the inventory
886
repository.add_inventory(rev.revision_id, inv, present_parents)
887
except errors.RevisionAlreadyPresent:
889
repository.add_revision(rev.revision_id, rev, inv)
892
class MetaDirRepository(Repository):
893
"""Repositories in the new meta-dir layout."""
895
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
896
super(MetaDirRepository, self).__init__(_format,
902
dir_mode = self.control_files._dir_mode
903
file_mode = self.control_files._file_mode
907
"""Return True if this repository is flagged as a shared repository."""
908
return self.control_files._transport.has('shared-storage')
911
def set_make_working_trees(self, new_value):
912
"""Set the policy flag for making working trees when creating branches.
914
This only applies to branches that use this repository.
916
The default is 'True'.
917
:param new_value: True to restore the default, False to disable making
922
self.control_files._transport.delete('no-working-trees')
923
except errors.NoSuchFile:
926
self.control_files.put_utf8('no-working-trees', '')
928
def make_working_trees(self):
929
"""Returns the policy for making working trees on new branches."""
930
return not self.control_files._transport.has('no-working-trees')
933
class WeaveMetaDirRepository(MetaDirRepository):
934
"""A subclass of MetaDirRepository to set weave specific policy."""
936
def get_commit_builder(self, branch, parents, config, timestamp=None,
937
timezone=None, committer=None, revprops=None,
939
self._check_ascii_revisionid(revision_id, self.get_commit_builder)
940
return MetaDirRepository.get_commit_builder(self, branch, parents,
941
config, timestamp, timezone, committer, revprops, revision_id)
944
class KnitRepository(MetaDirRepository):
945
"""Knit format repository."""
947
def _warn_if_deprecated(self):
948
# This class isn't deprecated
951
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
952
inv_vf.add_lines_with_ghosts(revid, parents, lines)
955
def _all_revision_ids(self):
956
"""See Repository.all_revision_ids()."""
957
# Knits get the revision graph from the index of the revision knit, so
958
# it's always possible even if they're on an unlistable transport.
959
return self._revision_store.all_revision_ids(self.get_transaction())
961
def fileid_involved_between_revs(self, from_revid, to_revid):
962
"""Find file_id(s) which are involved in the changes between revisions.
964
This determines the set of revisions which are involved, and then
965
finds all file ids affected by those revisions.
967
vf = self._get_revision_vf()
968
from_set = set(vf.get_ancestry(from_revid))
969
to_set = set(vf.get_ancestry(to_revid))
970
changed = to_set.difference(from_set)
971
return self._fileid_involved_by_set(changed)
973
def fileid_involved(self, last_revid=None):
974
"""Find all file_ids modified in the ancestry of last_revid.
976
:param last_revid: If None, last_revision() will be used.
979
changed = set(self.all_revision_ids())
981
changed = set(self.get_ancestry(last_revid))
984
return self._fileid_involved_by_set(changed)
987
def get_ancestry(self, revision_id):
988
"""Return a list of revision-ids integrated by a revision.
990
This is topologically sorted.
992
if revision_id is None:
994
vf = self._get_revision_vf()
996
return [None] + vf.get_ancestry(revision_id)
997
except errors.RevisionNotPresent:
998
raise errors.NoSuchRevision(self, revision_id)
1001
def get_revision(self, revision_id):
1002
"""Return the Revision object for a named revision"""
1003
return self.get_revision_reconcile(revision_id)
1006
def get_revision_graph(self, revision_id=None):
1007
"""Return a dictionary containing the revision graph.
1009
:param revision_id: The revision_id to get a graph from. If None, then
1010
the entire revision graph is returned. This is a deprecated mode of
1011
operation and will be removed in the future.
1012
:return: a dictionary of revision_id->revision_parents_list.
1014
# special case NULL_REVISION
1015
if revision_id == _mod_revision.NULL_REVISION:
1017
a_weave = self._get_revision_vf()
1018
entire_graph = a_weave.get_graph()
1019
if revision_id is None:
1020
return a_weave.get_graph()
1021
elif revision_id not in a_weave:
1022
raise errors.NoSuchRevision(self, revision_id)
1024
# add what can be reached from revision_id
1026
pending = set([revision_id])
1027
while len(pending) > 0:
1028
node = pending.pop()
1029
result[node] = a_weave.get_parents(node)
1030
for revision_id in result[node]:
1031
if revision_id not in result:
1032
pending.add(revision_id)
1036
def get_revision_graph_with_ghosts(self, revision_ids=None):
1037
"""Return a graph of the revisions with ghosts marked as applicable.
1039
:param revision_ids: an iterable of revisions to graph or None for all.
1040
:return: a Graph object with the graph reachable from revision_ids.
1042
result = graph.Graph()
1043
vf = self._get_revision_vf()
1044
versions = set(vf.versions())
1045
if not revision_ids:
1046
pending = set(self.all_revision_ids())
1049
pending = set(revision_ids)
1050
# special case NULL_REVISION
1051
if _mod_revision.NULL_REVISION in pending:
1052
pending.remove(_mod_revision.NULL_REVISION)
1053
required = set(pending)
1056
revision_id = pending.pop()
1057
if not revision_id in versions:
1058
if revision_id in required:
1059
raise errors.NoSuchRevision(self, revision_id)
1061
result.add_ghost(revision_id)
1062
# mark it as done so we don't try for it again.
1063
done.add(revision_id)
1065
parent_ids = vf.get_parents_with_ghosts(revision_id)
1066
for parent_id in parent_ids:
1067
# is this queued or done ?
1068
if (parent_id not in pending and
1069
parent_id not in done):
1071
pending.add(parent_id)
1072
result.add_node(revision_id, parent_ids)
1073
done.add(revision_id)
1076
def _get_revision_vf(self):
1077
""":return: a versioned file containing the revisions."""
1078
vf = self._revision_store.get_revision_file(self.get_transaction())
1082
def reconcile(self, other=None, thorough=False):
1083
"""Reconcile this repository."""
1084
from bzrlib.reconcile import KnitReconciler
1085
reconciler = KnitReconciler(self, thorough=thorough)
1086
reconciler.reconcile()
1089
def revision_parents(self, revision_id):
1090
return self._get_revision_vf().get_parents(revision_id)
1093
class KnitRepository2(KnitRepository):
1095
def __init__(self, _format, a_bzrdir, control_files, _revision_store,
1096
control_store, text_store):
1097
KnitRepository.__init__(self, _format, a_bzrdir, control_files,
1098
_revision_store, control_store, text_store)
1099
self._serializer = xml6.serializer_v6
1101
def deserialise_inventory(self, revision_id, xml):
1102
"""Transform the xml into an inventory object.
1104
:param revision_id: The expected revision id of the inventory.
1105
:param xml: A serialised inventory.
1107
result = self._serializer.read_inventory_from_string(xml)
1108
assert result.root.revision is not None
1111
def serialise_inventory(self, inv):
1112
"""Transform the inventory object into XML text.
1114
:param revision_id: The expected revision id of the inventory.
1115
:param xml: A serialised inventory.
1117
assert inv.revision_id is not None
1118
assert inv.root.revision is not None
1119
return KnitRepository.serialise_inventory(self, inv)
1121
def get_commit_builder(self, branch, parents, config, timestamp=None,
1122
timezone=None, committer=None, revprops=None,
1124
"""Obtain a CommitBuilder for this repository.
1126
:param branch: Branch to commit to.
1127
:param parents: Revision ids of the parents of the new revision.
1128
:param config: Configuration to use.
1129
:param timestamp: Optional timestamp recorded for commit.
1130
:param timezone: Optional timezone for timestamp.
1131
:param committer: Optional committer to set for commit.
1132
:param revprops: Optional dictionary of revision properties.
1133
:param revision_id: Optional revision id.
1135
return RootCommitBuilder(self, parents, config, timestamp, timezone,
1136
committer, revprops, revision_id)
1139
class RepositoryFormat(object):
1140
"""A repository format.
1142
Formats provide three things:
1143
* An initialization routine to construct repository data on disk.
1144
* a format string which is used when the BzrDir supports versioned
1146
* an open routine which returns a Repository instance.
1148
Formats are placed in an dict by their format string for reference
1149
during opening. These should be subclasses of RepositoryFormat
1152
Once a format is deprecated, just deprecate the initialize and open
1153
methods on the format class. Do not deprecate the object, as the
1154
object will be created every system load.
1156
Common instance attributes:
1157
_matchingbzrdir - the bzrdir format that the repository format was
1158
originally written to work with. This can be used if manually
1159
constructing a bzrdir and repository, or more commonly for test suite
1163
_registry = registry.Registry()
1164
"""Registry of formats, indexed by their identifying format string."""
1167
return "<%s>" % self.__class__.__name__
1170
def find_format(klass, a_bzrdir):
1171
"""Return the format for the repository object in a_bzrdir.
1173
This is used by bzr native formats that have a "format" file in
1174
the repository. Other methods may be used by different types of
1178
transport = a_bzrdir.get_repository_transport(None)
1179
format_string = transport.get("format").read()
1180
return klass._registry.get(format_string)
1181
except errors.NoSuchFile:
1182
raise errors.NoRepositoryPresent(a_bzrdir)
1184
raise errors.UnknownFormatError(format=format_string)
1187
def register_format(klass, format):
1188
klass._registry.register(format.get_format_string(), format)
1191
@deprecated_method(symbol_versioning.zero_fourteen)
1192
def set_default_format(klass, format):
1193
klass._set_default_format(format)
1196
def _set_default_format(klass, format):
1197
"""Set the default format for new Repository creation.
1199
The format must already be registered.
1201
klass._registry.default_key = format.get_format_string()
1204
def unregister_format(klass, format):
1205
klass._registry.remove(format.get_format_string())
1208
def get_default_format(klass):
1209
"""Return the current default format."""
1210
return klass._registry.get(klass._registry.default_key)
1212
def _get_control_store(self, repo_transport, control_files):
1213
"""Return the control store for this repository."""
1214
raise NotImplementedError(self._get_control_store)
1216
def get_format_string(self):
1217
"""Return the ASCII format string that identifies this format.
1219
Note that in pre format ?? repositories the format string is
1220
not permitted nor written to disk.
1222
raise NotImplementedError(self.get_format_string)
1224
def get_format_description(self):
1225
"""Return the short description for this format."""
1226
raise NotImplementedError(self.get_format_description)
1228
def _get_revision_store(self, repo_transport, control_files):
1229
"""Return the revision store object for this a_bzrdir."""
1230
raise NotImplementedError(self._get_revision_store)
1232
def _get_text_rev_store(self,
1239
"""Common logic for getting a revision store for a repository.
1241
see self._get_revision_store for the subclass-overridable method to
1242
get the store for a repository.
1244
from bzrlib.store.revision.text import TextRevisionStore
1245
dir_mode = control_files._dir_mode
1246
file_mode = control_files._file_mode
1247
text_store =TextStore(transport.clone(name),
1249
compressed=compressed,
1251
file_mode=file_mode)
1252
_revision_store = TextRevisionStore(text_store, serializer)
1253
return _revision_store
1255
def _get_versioned_file_store(self,
1260
versionedfile_class=weave.WeaveFile,
1261
versionedfile_kwargs={},
1263
weave_transport = control_files._transport.clone(name)
1264
dir_mode = control_files._dir_mode
1265
file_mode = control_files._file_mode
1266
return VersionedFileStore(weave_transport, prefixed=prefixed,
1268
file_mode=file_mode,
1269
versionedfile_class=versionedfile_class,
1270
versionedfile_kwargs=versionedfile_kwargs,
1273
def initialize(self, a_bzrdir, shared=False):
1274
"""Initialize a repository of this format in a_bzrdir.
1276
:param a_bzrdir: The bzrdir to put the new repository in it.
1277
:param shared: The repository should be initialized as a sharable one.
1279
This may raise UninitializableFormat if shared repository are not
1280
compatible the a_bzrdir.
1283
def is_supported(self):
1284
"""Is this format supported?
1286
Supported formats must be initializable and openable.
1287
Unsupported formats may not support initialization or committing or
1288
some other features depending on the reason for not being supported.
1292
def check_conversion_target(self, target_format):
1293
raise NotImplementedError(self.check_conversion_target)
1295
def open(self, a_bzrdir, _found=False):
1296
"""Return an instance of this format for the bzrdir a_bzrdir.
1298
_found is a private parameter, do not use it.
1300
raise NotImplementedError(self.open)
1303
class PreSplitOutRepositoryFormat(RepositoryFormat):
1304
"""Base class for the pre split out repository formats."""
1306
rich_root_data = False
1308
def initialize(self, a_bzrdir, shared=False, _internal=False):
1309
"""Create a weave repository.
1311
TODO: when creating split out bzr branch formats, move this to a common
1312
base for Format5, Format6. or something like that.
1315
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1318
# always initialized when the bzrdir is.
1319
return self.open(a_bzrdir, _found=True)
1321
# Create an empty weave
1323
weavefile.write_weave_v5(weave.Weave(), sio)
1324
empty_weave = sio.getvalue()
1326
mutter('creating repository in %s.', a_bzrdir.transport.base)
1327
dirs = ['revision-store', 'weaves']
1328
files = [('inventory.weave', StringIO(empty_weave)),
1331
# FIXME: RBC 20060125 don't peek under the covers
1332
# NB: no need to escape relative paths that are url safe.
1333
control_files = lockable_files.LockableFiles(a_bzrdir.transport,
1334
'branch-lock', lockable_files.TransportLock)
1335
control_files.create_lock()
1336
control_files.lock_write()
1337
control_files._transport.mkdir_multi(dirs,
1338
mode=control_files._dir_mode)
1340
for file, content in files:
1341
control_files.put(file, content)
1343
control_files.unlock()
1344
return self.open(a_bzrdir, _found=True)
1346
def _get_control_store(self, repo_transport, control_files):
1347
"""Return the control store for this repository."""
1348
return self._get_versioned_file_store('',
1353
def _get_text_store(self, transport, control_files):
1354
"""Get a store for file texts for this format."""
1355
raise NotImplementedError(self._get_text_store)
1357
def open(self, a_bzrdir, _found=False):
1358
"""See RepositoryFormat.open()."""
1360
# we are being called directly and must probe.
1361
raise NotImplementedError
1363
repo_transport = a_bzrdir.get_repository_transport(None)
1364
control_files = a_bzrdir._control_files
1365
text_store = self._get_text_store(repo_transport, control_files)
1366
control_store = self._get_control_store(repo_transport, control_files)
1367
_revision_store = self._get_revision_store(repo_transport, control_files)
1368
return AllInOneRepository(_format=self,
1370
_revision_store=_revision_store,
1371
control_store=control_store,
1372
text_store=text_store)
1374
def check_conversion_target(self, target_format):
1378
class RepositoryFormat4(PreSplitOutRepositoryFormat):
1379
"""Bzr repository format 4.
1381
This repository format has:
1383
- TextStores for texts, inventories,revisions.
1385
This format is deprecated: it indexes texts using a text id which is
1386
removed in format 5; initialization and write support for this format
1391
super(RepositoryFormat4, self).__init__()
1392
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1394
def get_format_description(self):
1395
"""See RepositoryFormat.get_format_description()."""
1396
return "Repository format 4"
1398
def initialize(self, url, shared=False, _internal=False):
1399
"""Format 4 branches cannot be created."""
1400
raise errors.UninitializableFormat(self)
1402
def is_supported(self):
1403
"""Format 4 is not supported.
1405
It is not supported because the model changed from 4 to 5 and the
1406
conversion logic is expensive - so doing it on the fly was not
1411
def _get_control_store(self, repo_transport, control_files):
1412
"""Format 4 repositories have no formal control store at this point.
1414
This will cause any control-file-needing apis to fail - this is desired.
1418
def _get_revision_store(self, repo_transport, control_files):
1419
"""See RepositoryFormat._get_revision_store()."""
1420
from bzrlib.xml4 import serializer_v4
1421
return self._get_text_rev_store(repo_transport,
1424
serializer=serializer_v4)
1426
def _get_text_store(self, transport, control_files):
1427
"""See RepositoryFormat._get_text_store()."""
1430
class RepositoryFormat5(PreSplitOutRepositoryFormat):
1431
"""Bzr control format 5.
1433
This repository format has:
1434
- weaves for file texts and inventory
1436
- TextStores for revisions and signatures.
1440
super(RepositoryFormat5, self).__init__()
1441
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1443
def get_format_description(self):
1444
"""See RepositoryFormat.get_format_description()."""
1445
return "Weave repository format 5"
1447
def _get_revision_store(self, repo_transport, control_files):
1448
"""See RepositoryFormat._get_revision_store()."""
1449
"""Return the revision store object for this a_bzrdir."""
1450
return self._get_text_rev_store(repo_transport,
1455
def _get_text_store(self, transport, control_files):
1456
"""See RepositoryFormat._get_text_store()."""
1457
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1460
class RepositoryFormat6(PreSplitOutRepositoryFormat):
1461
"""Bzr control format 6.
1463
This repository format has:
1464
- weaves for file texts and inventory
1465
- hash subdirectory based stores.
1466
- TextStores for revisions and signatures.
1470
super(RepositoryFormat6, self).__init__()
1471
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1473
def get_format_description(self):
1474
"""See RepositoryFormat.get_format_description()."""
1475
return "Weave repository format 6"
1477
def _get_revision_store(self, repo_transport, control_files):
1478
"""See RepositoryFormat._get_revision_store()."""
1479
return self._get_text_rev_store(repo_transport,
1485
def _get_text_store(self, transport, control_files):
1486
"""See RepositoryFormat._get_text_store()."""
1487
return self._get_versioned_file_store('weaves', transport, control_files)
1490
class MetaDirRepositoryFormat(RepositoryFormat):
1491
"""Common base class for the new repositories using the metadir layout."""
1493
rich_root_data = False
1496
super(MetaDirRepositoryFormat, self).__init__()
1497
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1499
def _create_control_files(self, a_bzrdir):
1500
"""Create the required files and the initial control_files object."""
1501
# FIXME: RBC 20060125 don't peek under the covers
1502
# NB: no need to escape relative paths that are url safe.
1503
repository_transport = a_bzrdir.get_repository_transport(self)
1504
control_files = lockable_files.LockableFiles(repository_transport,
1505
'lock', lockdir.LockDir)
1506
control_files.create_lock()
1507
return control_files
1509
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1510
"""Upload the initial blank content."""
1511
control_files = self._create_control_files(a_bzrdir)
1512
control_files.lock_write()
1514
control_files._transport.mkdir_multi(dirs,
1515
mode=control_files._dir_mode)
1516
for file, content in files:
1517
control_files.put(file, content)
1518
for file, content in utf8_files:
1519
control_files.put_utf8(file, content)
1521
control_files.put_utf8('shared-storage', '')
1523
control_files.unlock()
1526
class RepositoryFormat7(MetaDirRepositoryFormat):
1527
"""Bzr repository 7.
1529
This repository format has:
1530
- weaves for file texts and inventory
1531
- hash subdirectory based stores.
1532
- TextStores for revisions and signatures.
1533
- a format marker of its own
1534
- an optional 'shared-storage' flag
1535
- an optional 'no-working-trees' flag
1538
def _get_control_store(self, repo_transport, control_files):
1539
"""Return the control store for this repository."""
1540
return self._get_versioned_file_store('',
1545
def get_format_string(self):
1546
"""See RepositoryFormat.get_format_string()."""
1547
return "Bazaar-NG Repository format 7"
1549
def get_format_description(self):
1550
"""See RepositoryFormat.get_format_description()."""
1551
return "Weave repository format 7"
1553
def check_conversion_target(self, target_format):
1556
def _get_revision_store(self, repo_transport, control_files):
1557
"""See RepositoryFormat._get_revision_store()."""
1558
return self._get_text_rev_store(repo_transport,
1565
def _get_text_store(self, transport, control_files):
1566
"""See RepositoryFormat._get_text_store()."""
1567
return self._get_versioned_file_store('weaves',
1571
def initialize(self, a_bzrdir, shared=False):
1572
"""Create a weave repository.
1574
:param shared: If true the repository will be initialized as a shared
1577
# Create an empty weave
1579
weavefile.write_weave_v5(weave.Weave(), sio)
1580
empty_weave = sio.getvalue()
1582
mutter('creating repository in %s.', a_bzrdir.transport.base)
1583
dirs = ['revision-store', 'weaves']
1584
files = [('inventory.weave', StringIO(empty_weave)),
1586
utf8_files = [('format', self.get_format_string())]
1588
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1589
return self.open(a_bzrdir=a_bzrdir, _found=True)
1591
def open(self, a_bzrdir, _found=False, _override_transport=None):
1592
"""See RepositoryFormat.open().
1594
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1595
repository at a slightly different url
1596
than normal. I.e. during 'upgrade'.
1599
format = RepositoryFormat.find_format(a_bzrdir)
1600
assert format.__class__ == self.__class__
1601
if _override_transport is not None:
1602
repo_transport = _override_transport
1604
repo_transport = a_bzrdir.get_repository_transport(None)
1605
control_files = lockable_files.LockableFiles(repo_transport,
1606
'lock', lockdir.LockDir)
1607
text_store = self._get_text_store(repo_transport, control_files)
1608
control_store = self._get_control_store(repo_transport, control_files)
1609
_revision_store = self._get_revision_store(repo_transport, control_files)
1610
return WeaveMetaDirRepository(_format=self,
1612
control_files=control_files,
1613
_revision_store=_revision_store,
1614
control_store=control_store,
1615
text_store=text_store)
1618
class RepositoryFormatKnit(MetaDirRepositoryFormat):
1619
"""Bzr repository knit format (generalized).
1621
This repository format has:
1622
- knits for file texts and inventory
1623
- hash subdirectory based stores.
1624
- knits for revisions and signatures
1625
- TextStores for revisions and signatures.
1626
- a format marker of its own
1627
- an optional 'shared-storage' flag
1628
- an optional 'no-working-trees' flag
1632
def _get_control_store(self, repo_transport, control_files):
1633
"""Return the control store for this repository."""
1634
return VersionedFileStore(
1637
file_mode=control_files._file_mode,
1638
versionedfile_class=knit.KnitVersionedFile,
1639
versionedfile_kwargs={'factory':knit.KnitPlainFactory()},
1642
def _get_revision_store(self, repo_transport, control_files):
1643
"""See RepositoryFormat._get_revision_store()."""
1644
from bzrlib.store.revision.knit import KnitRevisionStore
1645
versioned_file_store = VersionedFileStore(
1647
file_mode=control_files._file_mode,
1650
versionedfile_class=knit.KnitVersionedFile,
1651
versionedfile_kwargs={'delta':False,
1652
'factory':knit.KnitPlainFactory(),
1656
return KnitRevisionStore(versioned_file_store)
1658
def _get_text_store(self, transport, control_files):
1659
"""See RepositoryFormat._get_text_store()."""
1660
return self._get_versioned_file_store('knits',
1663
versionedfile_class=knit.KnitVersionedFile,
1664
versionedfile_kwargs={
1665
'create_parent_dir':True,
1666
'delay_create':True,
1667
'dir_mode':control_files._dir_mode,
1671
def initialize(self, a_bzrdir, shared=False):
1672
"""Create a knit format 1 repository.
1674
:param a_bzrdir: bzrdir to contain the new repository; must already
1676
:param shared: If true the repository will be initialized as a shared
1679
mutter('creating repository in %s.', a_bzrdir.transport.base)
1680
dirs = ['revision-store', 'knits']
1682
utf8_files = [('format', self.get_format_string())]
1684
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1685
repo_transport = a_bzrdir.get_repository_transport(None)
1686
control_files = lockable_files.LockableFiles(repo_transport,
1687
'lock', lockdir.LockDir)
1688
control_store = self._get_control_store(repo_transport, control_files)
1689
transaction = transactions.WriteTransaction()
1690
# trigger a write of the inventory store.
1691
control_store.get_weave_or_empty('inventory', transaction)
1692
_revision_store = self._get_revision_store(repo_transport, control_files)
1693
# the revision id here is irrelevant: it will not be stored, and cannot
1695
_revision_store.has_revision_id('A', transaction)
1696
_revision_store.get_signature_file(transaction)
1697
return self.open(a_bzrdir=a_bzrdir, _found=True)
1699
def open(self, a_bzrdir, _found=False, _override_transport=None):
1700
"""See RepositoryFormat.open().
1702
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1703
repository at a slightly different url
1704
than normal. I.e. during 'upgrade'.
1707
format = RepositoryFormat.find_format(a_bzrdir)
1708
assert format.__class__ == self.__class__
1709
if _override_transport is not None:
1710
repo_transport = _override_transport
1712
repo_transport = a_bzrdir.get_repository_transport(None)
1713
control_files = lockable_files.LockableFiles(repo_transport,
1714
'lock', lockdir.LockDir)
1715
text_store = self._get_text_store(repo_transport, control_files)
1716
control_store = self._get_control_store(repo_transport, control_files)
1717
_revision_store = self._get_revision_store(repo_transport, control_files)
1718
return KnitRepository(_format=self,
1720
control_files=control_files,
1721
_revision_store=_revision_store,
1722
control_store=control_store,
1723
text_store=text_store)
1726
class RepositoryFormatKnit1(RepositoryFormatKnit):
1727
"""Bzr repository knit format 1.
1729
This repository format has:
1730
- knits for file texts and inventory
1731
- hash subdirectory based stores.
1732
- knits for revisions and signatures
1733
- TextStores for revisions and signatures.
1734
- a format marker of its own
1735
- an optional 'shared-storage' flag
1736
- an optional 'no-working-trees' flag
1739
This format was introduced in bzr 0.8.
1741
def get_format_string(self):
1742
"""See RepositoryFormat.get_format_string()."""
1743
return "Bazaar-NG Knit Repository Format 1"
1745
def get_format_description(self):
1746
"""See RepositoryFormat.get_format_description()."""
1747
return "Knit repository format 1"
1749
def check_conversion_target(self, target_format):
1753
class RepositoryFormatKnit2(RepositoryFormatKnit):
1754
"""Bzr repository knit format 2.
1756
THIS FORMAT IS EXPERIMENTAL
1757
This repository format has:
1758
- knits for file texts and inventory
1759
- hash subdirectory based stores.
1760
- knits for revisions and signatures
1761
- TextStores for revisions and signatures.
1762
- a format marker of its own
1763
- an optional 'shared-storage' flag
1764
- an optional 'no-working-trees' flag
1766
- Support for recording full info about the tree root
1770
rich_root_data = True
1772
def get_format_string(self):
1773
"""See RepositoryFormat.get_format_string()."""
1774
return "Bazaar Knit Repository Format 2\n"
1776
def get_format_description(self):
1777
"""See RepositoryFormat.get_format_description()."""
1778
return "Knit repository format 2"
1780
def check_conversion_target(self, target_format):
1781
if not target_format.rich_root_data:
1782
raise errors.BadConversionTarget(
1783
'Does not support rich root data.', target_format)
1785
def open(self, a_bzrdir, _found=False, _override_transport=None):
1786
"""See RepositoryFormat.open().
1788
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1789
repository at a slightly different url
1790
than normal. I.e. during 'upgrade'.
1793
format = RepositoryFormat.find_format(a_bzrdir)
1794
assert format.__class__ == self.__class__
1795
if _override_transport is not None:
1796
repo_transport = _override_transport
1798
repo_transport = a_bzrdir.get_repository_transport(None)
1799
control_files = lockable_files.LockableFiles(repo_transport, 'lock',
1801
text_store = self._get_text_store(repo_transport, control_files)
1802
control_store = self._get_control_store(repo_transport, control_files)
1803
_revision_store = self._get_revision_store(repo_transport, control_files)
1804
return KnitRepository2(_format=self,
1806
control_files=control_files,
1807
_revision_store=_revision_store,
1808
control_store=control_store,
1809
text_store=text_store)
1813
# formats which have no format string are not discoverable
1814
# and not independently creatable, so are not registered.
1815
RepositoryFormat.register_format(RepositoryFormat7())
1816
# KEEP in sync with bzrdir.format_registry default, which controls the overall
1817
# default control directory format
1818
_default_format = RepositoryFormatKnit1()
1819
RepositoryFormat.register_format(_default_format)
1820
RepositoryFormat.register_format(RepositoryFormatKnit2())
1821
RepositoryFormat._set_default_format(_default_format)
1822
_legacy_formats = [RepositoryFormat4(),
1823
RepositoryFormat5(),
1824
RepositoryFormat6()]
1827
class InterRepository(InterObject):
1828
"""This class represents operations taking place between two repositories.
1830
Its instances have methods like copy_content and fetch, and contain
1831
references to the source and target repositories these operations can be
1834
Often we will provide convenience methods on 'repository' which carry out
1835
operations with another repository - they will always forward to
1836
InterRepository.get(other).method_name(parameters).
1840
"""The available optimised InterRepository types."""
1842
def copy_content(self, revision_id=None, basis=None):
1843
raise NotImplementedError(self.copy_content)
1845
def fetch(self, revision_id=None, pb=None):
1846
"""Fetch the content required to construct revision_id.
1848
The content is copied from self.source to self.target.
1850
:param revision_id: if None all content is copied, if NULL_REVISION no
1852
:param pb: optional progress bar to use for progress reports. If not
1853
provided a default one will be created.
1855
Returns the copied revision count and the failed revisions in a tuple:
1858
raise NotImplementedError(self.fetch)
1861
def missing_revision_ids(self, revision_id=None):
1862
"""Return the revision ids that source has that target does not.
1864
These are returned in topological order.
1866
:param revision_id: only return revision ids included by this
1869
# generic, possibly worst case, slow code path.
1870
target_ids = set(self.target.all_revision_ids())
1871
if revision_id is not None:
1872
source_ids = self.source.get_ancestry(revision_id)
1873
assert source_ids[0] is None
1876
source_ids = self.source.all_revision_ids()
1877
result_set = set(source_ids).difference(target_ids)
1878
# this may look like a no-op: its not. It preserves the ordering
1879
# other_ids had while only returning the members from other_ids
1880
# that we've decided we need.
1881
return [rev_id for rev_id in source_ids if rev_id in result_set]
1884
class InterSameDataRepository(InterRepository):
1885
"""Code for converting between repositories that represent the same data.
1887
Data format and model must match for this to work.
1890
_matching_repo_format = RepositoryFormat4()
1891
"""Repository format for testing with."""
1894
def is_compatible(source, target):
1895
if not isinstance(source, Repository):
1897
if not isinstance(target, Repository):
1899
if source._format.rich_root_data == target._format.rich_root_data:
1905
def copy_content(self, revision_id=None, basis=None):
1906
"""Make a complete copy of the content in self into destination.
1908
This is a destructive operation! Do not use it on existing
1911
:param revision_id: Only copy the content needed to construct
1912
revision_id and its parents.
1913
:param basis: Copy the needed data preferentially from basis.
1916
self.target.set_make_working_trees(self.source.make_working_trees())
1917
except NotImplementedError:
1919
# grab the basis available data
1920
if basis is not None:
1921
self.target.fetch(basis, revision_id=revision_id)
1922
# but don't bother fetching if we have the needed data now.
1923
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
1924
self.target.has_revision(revision_id)):
1926
self.target.fetch(self.source, revision_id=revision_id)
1929
def fetch(self, revision_id=None, pb=None):
1930
"""See InterRepository.fetch()."""
1931
from bzrlib.fetch import GenericRepoFetcher
1932
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1933
self.source, self.source._format, self.target,
1934
self.target._format)
1935
f = GenericRepoFetcher(to_repository=self.target,
1936
from_repository=self.source,
1937
last_revision=revision_id,
1939
return f.count_copied, f.failed_revisions
1942
class InterWeaveRepo(InterSameDataRepository):
1943
"""Optimised code paths between Weave based repositories."""
1945
_matching_repo_format = RepositoryFormat7()
1946
"""Repository format for testing with."""
1949
def is_compatible(source, target):
1950
"""Be compatible with known Weave formats.
1952
We don't test for the stores being of specific types because that
1953
could lead to confusing results, and there is no need to be
1957
return (isinstance(source._format, (RepositoryFormat5,
1959
RepositoryFormat7)) and
1960
isinstance(target._format, (RepositoryFormat5,
1962
RepositoryFormat7)))
1963
except AttributeError:
1967
def copy_content(self, revision_id=None, basis=None):
1968
"""See InterRepository.copy_content()."""
1969
# weave specific optimised path:
1970
if basis is not None:
1971
# copy the basis in, then fetch remaining data.
1972
basis.copy_content_into(self.target, revision_id)
1973
# the basis copy_content_into could miss-set this.
1975
self.target.set_make_working_trees(self.source.make_working_trees())
1976
except NotImplementedError:
1978
self.target.fetch(self.source, revision_id=revision_id)
1981
self.target.set_make_working_trees(self.source.make_working_trees())
1982
except NotImplementedError:
1984
# FIXME do not peek!
1985
if self.source.control_files._transport.listable():
1986
pb = ui.ui_factory.nested_progress_bar()
1988
self.target.weave_store.copy_all_ids(
1989
self.source.weave_store,
1991
from_transaction=self.source.get_transaction(),
1992
to_transaction=self.target.get_transaction())
1993
pb.update('copying inventory', 0, 1)
1994
self.target.control_weaves.copy_multi(
1995
self.source.control_weaves, ['inventory'],
1996
from_transaction=self.source.get_transaction(),
1997
to_transaction=self.target.get_transaction())
1998
self.target._revision_store.text_store.copy_all_ids(
1999
self.source._revision_store.text_store,
2004
self.target.fetch(self.source, revision_id=revision_id)
2007
def fetch(self, revision_id=None, pb=None):
2008
"""See InterRepository.fetch()."""
2009
from bzrlib.fetch import GenericRepoFetcher
2010
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2011
self.source, self.source._format, self.target, self.target._format)
2012
f = GenericRepoFetcher(to_repository=self.target,
2013
from_repository=self.source,
2014
last_revision=revision_id,
2016
return f.count_copied, f.failed_revisions
2019
def missing_revision_ids(self, revision_id=None):
2020
"""See InterRepository.missing_revision_ids()."""
2021
# we want all revisions to satisfy revision_id in source.
2022
# but we don't want to stat every file here and there.
2023
# we want then, all revisions other needs to satisfy revision_id
2024
# checked, but not those that we have locally.
2025
# so the first thing is to get a subset of the revisions to
2026
# satisfy revision_id in source, and then eliminate those that
2027
# we do already have.
2028
# this is slow on high latency connection to self, but as as this
2029
# disk format scales terribly for push anyway due to rewriting
2030
# inventory.weave, this is considered acceptable.
2032
if revision_id is not None:
2033
source_ids = self.source.get_ancestry(revision_id)
2034
assert source_ids[0] is None
2037
source_ids = self.source._all_possible_ids()
2038
source_ids_set = set(source_ids)
2039
# source_ids is the worst possible case we may need to pull.
2040
# now we want to filter source_ids against what we actually
2041
# have in target, but don't try to check for existence where we know
2042
# we do not have a revision as that would be pointless.
2043
target_ids = set(self.target._all_possible_ids())
2044
possibly_present_revisions = target_ids.intersection(source_ids_set)
2045
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2046
required_revisions = source_ids_set.difference(actually_present_revisions)
2047
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2048
if revision_id is not None:
2049
# we used get_ancestry to determine source_ids then we are assured all
2050
# revisions referenced are present as they are installed in topological order.
2051
# and the tip revision was validated by get_ancestry.
2052
return required_topo_revisions
2054
# if we just grabbed the possibly available ids, then
2055
# we only have an estimate of whats available and need to validate
2056
# that against the revision records.
2057
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2060
class InterKnitRepo(InterSameDataRepository):
2061
"""Optimised code paths between Knit based repositories."""
2063
_matching_repo_format = RepositoryFormatKnit1()
2064
"""Repository format for testing with."""
2067
def is_compatible(source, target):
2068
"""Be compatible with known Knit formats.
2070
We don't test for the stores being of specific types because that
2071
could lead to confusing results, and there is no need to be
2075
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2076
isinstance(target._format, (RepositoryFormatKnit1)))
2077
except AttributeError:
2081
def fetch(self, revision_id=None, pb=None):
2082
"""See InterRepository.fetch()."""
2083
from bzrlib.fetch import KnitRepoFetcher
2084
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2085
self.source, self.source._format, self.target, self.target._format)
2086
f = KnitRepoFetcher(to_repository=self.target,
2087
from_repository=self.source,
2088
last_revision=revision_id,
2090
return f.count_copied, f.failed_revisions
2093
def missing_revision_ids(self, revision_id=None):
2094
"""See InterRepository.missing_revision_ids()."""
2095
if revision_id is not None:
2096
source_ids = self.source.get_ancestry(revision_id)
2097
assert source_ids[0] is None
2100
source_ids = self.source._all_possible_ids()
2101
source_ids_set = set(source_ids)
2102
# source_ids is the worst possible case we may need to pull.
2103
# now we want to filter source_ids against what we actually
2104
# have in target, but don't try to check for existence where we know
2105
# we do not have a revision as that would be pointless.
2106
target_ids = set(self.target._all_possible_ids())
2107
possibly_present_revisions = target_ids.intersection(source_ids_set)
2108
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
2109
required_revisions = source_ids_set.difference(actually_present_revisions)
2110
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
2111
if revision_id is not None:
2112
# we used get_ancestry to determine source_ids then we are assured all
2113
# revisions referenced are present as they are installed in topological order.
2114
# and the tip revision was validated by get_ancestry.
2115
return required_topo_revisions
2117
# if we just grabbed the possibly available ids, then
2118
# we only have an estimate of whats available and need to validate
2119
# that against the revision records.
2120
return self.source._eliminate_revisions_not_present(required_topo_revisions)
2123
class InterModel1and2(InterRepository):
2125
_matching_repo_format = None
2128
def is_compatible(source, target):
2129
if not isinstance(source, Repository):
2131
if not isinstance(target, Repository):
2133
if not source._format.rich_root_data and target._format.rich_root_data:
2139
def fetch(self, revision_id=None, pb=None):
2140
"""See InterRepository.fetch()."""
2141
from bzrlib.fetch import Model1toKnit2Fetcher
2142
f = Model1toKnit2Fetcher(to_repository=self.target,
2143
from_repository=self.source,
2144
last_revision=revision_id,
2146
return f.count_copied, f.failed_revisions
2149
def copy_content(self, revision_id=None, basis=None):
2150
"""Make a complete copy of the content in self into destination.
2152
This is a destructive operation! Do not use it on existing
2155
:param revision_id: Only copy the content needed to construct
2156
revision_id and its parents.
2157
:param basis: Copy the needed data preferentially from basis.
2160
self.target.set_make_working_trees(self.source.make_working_trees())
2161
except NotImplementedError:
2163
# grab the basis available data
2164
if basis is not None:
2165
self.target.fetch(basis, revision_id=revision_id)
2166
# but don't bother fetching if we have the needed data now.
2167
if (revision_id not in (None, _mod_revision.NULL_REVISION) and
2168
self.target.has_revision(revision_id)):
2170
self.target.fetch(self.source, revision_id=revision_id)
2173
class InterKnit1and2(InterKnitRepo):
2175
_matching_repo_format = None
2178
def is_compatible(source, target):
2179
"""Be compatible with Knit1 source and Knit2 target"""
2181
return (isinstance(source._format, (RepositoryFormatKnit1)) and
2182
isinstance(target._format, (RepositoryFormatKnit2)))
2183
except AttributeError:
2187
def fetch(self, revision_id=None, pb=None):
2188
"""See InterRepository.fetch()."""
2189
from bzrlib.fetch import Knit1to2Fetcher
2190
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
2191
self.source, self.source._format, self.target,
2192
self.target._format)
2193
f = Knit1to2Fetcher(to_repository=self.target,
2194
from_repository=self.source,
2195
last_revision=revision_id,
2197
return f.count_copied, f.failed_revisions
2200
InterRepository.register_optimiser(InterSameDataRepository)
2201
InterRepository.register_optimiser(InterWeaveRepo)
2202
InterRepository.register_optimiser(InterKnitRepo)
2203
InterRepository.register_optimiser(InterModel1and2)
2204
InterRepository.register_optimiser(InterKnit1and2)
2207
class RepositoryTestProviderAdapter(object):
2208
"""A tool to generate a suite testing multiple repository formats at once.
2210
This is done by copying the test once for each transport and injecting
2211
the transport_server, transport_readonly_server, and bzrdir_format and
2212
repository_format classes into each copy. Each copy is also given a new id()
2213
to make it easy to identify.
2216
def __init__(self, transport_server, transport_readonly_server, formats):
2217
self._transport_server = transport_server
2218
self._transport_readonly_server = transport_readonly_server
2219
self._formats = formats
2221
def adapt(self, test):
2222
result = unittest.TestSuite()
2223
for repository_format, bzrdir_format in self._formats:
2224
new_test = deepcopy(test)
2225
new_test.transport_server = self._transport_server
2226
new_test.transport_readonly_server = self._transport_readonly_server
2227
new_test.bzrdir_format = bzrdir_format
2228
new_test.repository_format = repository_format
2229
def make_new_test_id():
2230
new_id = "%s(%s)" % (new_test.id(), repository_format.__class__.__name__)
2231
return lambda: new_id
2232
new_test.id = make_new_test_id()
2233
result.addTest(new_test)
2237
class InterRepositoryTestProviderAdapter(object):
2238
"""A tool to generate a suite testing multiple inter repository formats.
2240
This is done by copying the test once for each interrepo provider and injecting
2241
the transport_server, transport_readonly_server, repository_format and
2242
repository_to_format classes into each copy.
2243
Each copy is also given a new id() to make it easy to identify.
2246
def __init__(self, transport_server, transport_readonly_server, formats):
2247
self._transport_server = transport_server
2248
self._transport_readonly_server = transport_readonly_server
2249
self._formats = formats
2251
def adapt(self, test):
2252
result = unittest.TestSuite()
2253
for interrepo_class, repository_format, repository_format_to in self._formats:
2254
new_test = deepcopy(test)
2255
new_test.transport_server = self._transport_server
2256
new_test.transport_readonly_server = self._transport_readonly_server
2257
new_test.interrepo_class = interrepo_class
2258
new_test.repository_format = repository_format
2259
new_test.repository_format_to = repository_format_to
2260
def make_new_test_id():
2261
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
2262
return lambda: new_id
2263
new_test.id = make_new_test_id()
2264
result.addTest(new_test)
2268
def default_test_list():
2269
"""Generate the default list of interrepo permutations to test."""
2271
# test the default InterRepository between format 6 and the current
2273
# XXX: robertc 20060220 reinstate this when there are two supported
2274
# formats which do not have an optimal code path between them.
2275
#result.append((InterRepository,
2276
# RepositoryFormat6(),
2277
# RepositoryFormatKnit1()))
2278
for optimiser in InterRepository._optimisers:
2279
if optimiser._matching_repo_format is not None:
2280
result.append((optimiser,
2281
optimiser._matching_repo_format,
2282
optimiser._matching_repo_format
2284
# if there are specific combinations we want to use, we can add them
2286
result.append((InterModel1and2, RepositoryFormat5(),
2287
RepositoryFormatKnit2()))
2288
result.append((InterKnit1and2, RepositoryFormatKnit1(),
2289
RepositoryFormatKnit2()))
2293
class CopyConverter(object):
2294
"""A repository conversion tool which just performs a copy of the content.
2296
This is slow but quite reliable.
2299
def __init__(self, target_format):
2300
"""Create a CopyConverter.
2302
:param target_format: The format the resulting repository should be.
2304
self.target_format = target_format
2306
def convert(self, repo, pb):
2307
"""Perform the conversion of to_convert, giving feedback via pb.
2309
:param to_convert: The disk object to convert.
2310
:param pb: a progress bar to use for progress information.
2315
# this is only useful with metadir layouts - separated repo content.
2316
# trigger an assertion if not such
2317
repo._format.get_format_string()
2318
self.repo_dir = repo.bzrdir
2319
self.step('Moving repository to repository.backup')
2320
self.repo_dir.transport.move('repository', 'repository.backup')
2321
backup_transport = self.repo_dir.transport.clone('repository.backup')
2322
repo._format.check_conversion_target(self.target_format)
2323
self.source_repo = repo._format.open(self.repo_dir,
2325
_override_transport=backup_transport)
2326
self.step('Creating new repository')
2327
converted = self.target_format.initialize(self.repo_dir,
2328
self.source_repo.is_shared())
2329
converted.lock_write()
2331
self.step('Copying content into repository.')
2332
self.source_repo.copy_content_into(converted)
2335
self.step('Deleting old repository content.')
2336
self.repo_dir.transport.delete_tree('repository.backup')
2337
self.pb.note('repository converted')
2339
def step(self, message):
2340
"""Update the pb by a step."""
2342
self.pb.update(message, self.count, self.total)
2345
class CommitBuilder(object):
2346
"""Provides an interface to build up a commit.
2348
This allows describing a tree to be committed without needing to
2349
know the internals of the format of the repository.
2352
record_root_entry = False
2353
def __init__(self, repository, parents, config, timestamp=None,
2354
timezone=None, committer=None, revprops=None,
2356
"""Initiate a CommitBuilder.
2358
:param repository: Repository to commit to.
2359
:param parents: Revision ids of the parents of the new revision.
2360
:param config: Configuration to use.
2361
:param timestamp: Optional timestamp recorded for commit.
2362
:param timezone: Optional timezone for timestamp.
2363
:param committer: Optional committer to set for commit.
2364
:param revprops: Optional dictionary of revision properties.
2365
:param revision_id: Optional revision id.
2367
self._config = config
2369
if committer is None:
2370
self._committer = self._config.username()
2372
assert isinstance(committer, basestring), type(committer)
2373
self._committer = committer
2375
self.new_inventory = Inventory(None)
2376
self._new_revision_id = revision_id
2377
self.parents = parents
2378
self.repository = repository
2381
if revprops is not None:
2382
self._revprops.update(revprops)
2384
if timestamp is None:
2385
timestamp = time.time()
2386
# Restrict resolution to 1ms
2387
self._timestamp = round(timestamp, 3)
2389
if timezone is None:
2390
self._timezone = local_time_offset()
2392
self._timezone = int(timezone)
2394
self._generate_revision_if_needed()
2396
def commit(self, message):
2397
"""Make the actual commit.
2399
:return: The revision id of the recorded revision.
2401
rev = _mod_revision.Revision(
2402
timestamp=self._timestamp,
2403
timezone=self._timezone,
2404
committer=self._committer,
2406
inventory_sha1=self.inv_sha1,
2407
revision_id=self._new_revision_id,
2408
properties=self._revprops)
2409
rev.parent_ids = self.parents
2410
self.repository.add_revision(self._new_revision_id, rev,
2411
self.new_inventory, self._config)
2412
return self._new_revision_id
2414
def revision_tree(self):
2415
"""Return the tree that was just committed.
2417
After calling commit() this can be called to get a RevisionTree
2418
representing the newly committed tree. This is preferred to
2419
calling Repository.revision_tree() because that may require
2420
deserializing the inventory, while we already have a copy in
2423
return RevisionTree(self.repository, self.new_inventory,
2424
self._new_revision_id)
2426
def finish_inventory(self):
2427
"""Tell the builder that the inventory is finished."""
2428
if self.new_inventory.root is None:
2429
symbol_versioning.warn('Root entry should be supplied to'
2430
' record_entry_contents, as of bzr 0.10.',
2431
DeprecationWarning, stacklevel=2)
2432
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2433
self.new_inventory.revision_id = self._new_revision_id
2434
self.inv_sha1 = self.repository.add_inventory(
2435
self._new_revision_id,
2440
def _gen_revision_id(self):
2441
"""Return new revision-id."""
2442
return generate_ids.gen_revision_id(self._config.username(),
2445
def _generate_revision_if_needed(self):
2446
"""Create a revision id if None was supplied.
2448
If the repository can not support user-specified revision ids
2449
they should override this function and raise CannotSetRevisionId
2450
if _new_revision_id is not None.
2452
:raises: CannotSetRevisionId
2454
if self._new_revision_id is None:
2455
self._new_revision_id = self._gen_revision_id()
2457
def record_entry_contents(self, ie, parent_invs, path, tree):
2458
"""Record the content of ie from tree into the commit if needed.
2460
Side effect: sets ie.revision when unchanged
2462
:param ie: An inventory entry present in the commit.
2463
:param parent_invs: The inventories of the parent revisions of the
2465
:param path: The path the entry is at in the tree.
2466
:param tree: The tree which contains this entry and should be used to
2469
if self.new_inventory.root is None and ie.parent_id is not None:
2470
symbol_versioning.warn('Root entry should be supplied to'
2471
' record_entry_contents, as of bzr 0.10.',
2472
DeprecationWarning, stacklevel=2)
2473
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2475
self.new_inventory.add(ie)
2477
# ie.revision is always None if the InventoryEntry is considered
2478
# for committing. ie.snapshot will record the correct revision
2479
# which may be the sole parent if it is untouched.
2480
if ie.revision is not None:
2483
# In this revision format, root entries have no knit or weave
2484
if ie is self.new_inventory.root:
2485
# When serializing out to disk and back in
2486
# root.revision is always _new_revision_id
2487
ie.revision = self._new_revision_id
2489
previous_entries = ie.find_previous_heads(
2491
self.repository.weave_store,
2492
self.repository.get_transaction())
2493
# we are creating a new revision for ie in the history store
2495
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2497
def modified_directory(self, file_id, file_parents):
2498
"""Record the presence of a symbolic link.
2500
:param file_id: The file_id of the link to record.
2501
:param file_parents: The per-file parent revision ids.
2503
self._add_text_to_weave(file_id, [], file_parents.keys())
2505
def modified_file_text(self, file_id, file_parents,
2506
get_content_byte_lines, text_sha1=None,
2508
"""Record the text of file file_id
2510
:param file_id: The file_id of the file to record the text of.
2511
:param file_parents: The per-file parent revision ids.
2512
:param get_content_byte_lines: A callable which will return the byte
2514
:param text_sha1: Optional SHA1 of the file contents.
2515
:param text_size: Optional size of the file contents.
2517
# mutter('storing text of file {%s} in revision {%s} into %r',
2518
# file_id, self._new_revision_id, self.repository.weave_store)
2519
# special case to avoid diffing on renames or
2521
if (len(file_parents) == 1
2522
and text_sha1 == file_parents.values()[0].text_sha1
2523
and text_size == file_parents.values()[0].text_size):
2524
previous_ie = file_parents.values()[0]
2525
versionedfile = self.repository.weave_store.get_weave(file_id,
2526
self.repository.get_transaction())
2527
versionedfile.clone_text(self._new_revision_id,
2528
previous_ie.revision, file_parents.keys())
2529
return text_sha1, text_size
2531
new_lines = get_content_byte_lines()
2532
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2533
# should return the SHA1 and size
2534
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2535
return osutils.sha_strings(new_lines), \
2536
sum(map(len, new_lines))
2538
def modified_link(self, file_id, file_parents, link_target):
2539
"""Record the presence of a symbolic link.
2541
:param file_id: The file_id of the link to record.
2542
:param file_parents: The per-file parent revision ids.
2543
:param link_target: Target location of this link.
2545
self._add_text_to_weave(file_id, [], file_parents.keys())
2547
def _add_text_to_weave(self, file_id, new_lines, parents):
2548
versionedfile = self.repository.weave_store.get_weave_or_empty(
2549
file_id, self.repository.get_transaction())
2550
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2551
versionedfile.clear_cache()
2554
class _CommitBuilder(CommitBuilder):
2555
"""Temporary class so old CommitBuilders are detected properly
2557
Note: CommitBuilder works whether or not root entry is recorded.
2560
record_root_entry = True
2563
class RootCommitBuilder(CommitBuilder):
2564
"""This commitbuilder actually records the root id"""
2566
record_root_entry = True
2568
def record_entry_contents(self, ie, parent_invs, path, tree):
2569
"""Record the content of ie from tree into the commit if needed.
2571
Side effect: sets ie.revision when unchanged
2573
:param ie: An inventory entry present in the commit.
2574
:param parent_invs: The inventories of the parent revisions of the
2576
:param path: The path the entry is at in the tree.
2577
:param tree: The tree which contains this entry and should be used to
2580
assert self.new_inventory.root is not None or ie.parent_id is None
2581
self.new_inventory.add(ie)
2583
# ie.revision is always None if the InventoryEntry is considered
2584
# for committing. ie.snapshot will record the correct revision
2585
# which may be the sole parent if it is untouched.
2586
if ie.revision is not None:
2589
previous_entries = ie.find_previous_heads(
2591
self.repository.weave_store,
2592
self.repository.get_transaction())
2593
# we are creating a new revision for ie in the history store
2595
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2607
def _unescaper(match, _map=_unescape_map):
2608
return _map[match.group(1)]
2614
def _unescape_xml(data):
2615
"""Unescape predefined XML entities in a string of data."""
2617
if _unescape_re is None:
2618
_unescape_re = re.compile('\&([^;]*);')
2619
return _unescape_re.sub(_unescaper, data)